1. Recent warming and changes of circulation in the North Atlantic - simulated with eddy-permitting & eddy-resolving models Robert Marsh, Beverly de Cuevas, Andrew Coward & Simon Josey (+ contributions by Graham Quartly, Lisa Redbourn, Julian Jacquin) Effects of Climate Change on the World's Oceans, Gijon, 19 - 23 May 2008

2. Outline 1.Observed Warming - brief background 2.OCCAM hindcasting - brief background 3.Patterns of surface warming since the 1980s, in the North Atlantic 4.Role of anomalous surface fluxes vs. anomalous ocean heat transport 5.Anomalous ocean heat transport: meridional overturning circulation (MOC) vs. “non-MOC” (gyres, eddies) 6.Summary & Future Developments Prologue: Impacts are important, but the ocean is not a passive participant, so we may also consider: “Effect of the World’s Oceans on Climate Change”

3. North Atlantic surface warming since 1985: satellite measurements (Pathfinder, AMSR) monthly anomalies (relative to means over hindcast 1985-2003 reference period) considered in three zones revealing strong interannual variability in the tropics, strong decadal variability in mid-latitudes … and warming trends

4. North Atlantic sub-surface warming since 1999 revealed by measurements from Argo floats Argo data in 10° zones used to estimate heat content in upper 1500 m For two zones in particular, spanning 50-70°N (subpolar gyre), heat content increases strongly, by around 10 21 J/year In global & longer term contexts: over 1955-98, warming of the upper 3km of the Atlantic accounts for ~44% of heat gain by the entire Earth System (atmosphere, oceans, soils, ice) - extrapolating from Levitus et al. (2005) [courtesy of Vladimir Ivchenko - see Ivchenko et al., 2006, GRL]

5. The Ocean Circulation & Climate Advanced Model (OCCAM, 1995-2007)  OCCAM projects provided huge amounts of data for extensive analysis of key processes in ocean circulation  Crowning achievement of OCCAM: hindcast of World Ocean at 1/12° resolution (< 10 km gridcell dimensions), truly resolving eddies equatorward of ~50°  OCCAM analysis highlights: Antarctic Circumpolar Current; Agulhas Retroflection; Indonesian Throughflow  OCCAM superceded by models in a new framework: Nucleus for European Modeling of the Ocean (NEMO)

6. OCCAM: A Killworth Legacy Peter Killworth (1946-2008)  Late 1980s: Peter was a FRAM pioneer, in particular formulating a scheme for the “free surface” (also a key feature of OCCAM)  In OCCAM context, he subsequently worked on the bottom boundary layer, eddy parameterizations, Rossby waves, and many other things

7. The World Ocean of OCCAM … is actually 2 models joined together at the Atlantic Equator & connected at the Bering Strait (Pacific-Arctic)  Subsequently results from Atlantic/Arctic sector today …

8. OCCAM: 1/4° & 1/12° simulations, 1985-2004 Compared to OC-4, OC-12 has: Narrower Boundary Currents (Gulf Stream, subpolar gyre) More Correct SST field (especially “Northwest Corner”) “Proper” Eddies & Eddy Fluxes (e.g., Lee et al. 2007) OC-4OC-12 e.g., January 1989 Sea Surface Temperature in N. Atlantic:

9. Confronting OCCAM with Observations  Recent changes in AMOC & Heat Transports (Marsh et al. 2005, GRL)  Only a few estimates of heat transport - how representative of recent changes? Observed oceanic heat transport across hydrographic sections 26ºN, Spain-Greenland (1 PW = 10 15 W)

10. Confronting OCCAM with Observations  OCCAM time series reveal considerable variability in ocean heat transport, encompassing recent estimates OCCAM Observed oceanic heat transport across hydrographic sections 26ºN, Spain-Greenland  Recent changes in AMOC & Heat Transports (Marsh et al. 2005, GRL) (1 PW = 10 15 W)

11. Recent Atlantic warming: Significantly non-zero SST change simulated in OC-4, observed by satellites (Pathfinder) & in situ (NOC) Composite changes - 1995-2003 mean minus 1985-94 mean  OC-4 simulates observed warming in Tropics, Mid-latitudes  Analysis of OC-4 confirms the extent to which SST increases are the surface expression of increasing Ocean Heat Content …

12. Significant trend in OC-4 Ocean Heat Content (OHC, GJ m -2 year -1 ), 1985-2003 almost entirely in upper 1500m (layer-mean 1ºC warming in mid-lats) rate of warming in good agreement with Argo estimates A3 AR7W A25

13. Point-wise correlation coefficients between annual-mean OHC and annual-mean SST Correlations persist to great depth at mid and high latitudes, also in the eastern Tropics Highly significant correlations (r > 0.9) coincide with regions of strongest SST increase in Mid-latitudes

14. Annual tendency in OC-4 Heat Content, Q tend (units PW = 10 15 W): Tropics & Mid-latitudes Tendencies generally positive - i.e., long-term warming Much of full-depth heating in upper few hundred metres Some anti-correlation, notably in late 1990s, between Tropics & Mid-latitudes - suggests role for meridional heat transport?

15. OC-4 Heat Budget: Mid-latitudes Q net (OC-4) ≈ Q net (NOC 1.1a), so model surface heat fluxes close to some of our best observations Advection supplies heat (> 0), Surface Fluxes remove heat (< 0) annual net heating clearly linked to annual Q adv anomalies, especially in late 1990s

16. Additional OCCAM hindcasts: Atlantic MOC NADW AABW Later model version (improved physics?) + Alternative forcing Higher resolution (1/12°) “Original”

17. Meridional heat transport, OCCAM & observations  Split Total as: MOC (warm upper branch; cold lower branch) non-MOC (gyres & eddies: cold western boundary current; warm eastern interior flow)  Most of tropical & subtropical heat transport due to MOC  Most of mid-latitude heat transport due to gyre & eddies  Most emphatically in OC-12

18. Annual-mean AMOC intensity  OC-4 (runs 202, 104) simulated fairly steady AMOC (Marsh et al. 2005)  OC-4 (run 103) in contrast simulate AMOC decline (as Bryden et al. 2005, for 1990s)  OC-12 in further contrast simulates fairly steady AMOC  Demonstrating the difficulty of simulating AMOC changes!  Nevertheless, we can investigate variability of heat transport …

19. Monthly MHT anomalies  South of ~40°N: pre-dominantly interannual; wind forced?  North of ~40°N: pre-dominantly decadal; wind + buoyancy forced?

20. Monthly MHT anomalies - MOC part  MOC dominates heat transport variations south of ~40°N  Hence the RAPID monitoring array at 26°N  MOC has little influence on the decadal variability further north …

21. Monthly MHT anomalies - non-MOC part  … which are in fact associated with changes in the subpolar gyre and/or eddy fluxes

22. MHT: standard deviations & trends  Confirmation that variability is primarily associated with MOC component in tropics & sub-tropics  Variability associated with non-MOC component most prominent at inter-gyre boundary (eddies?)  Most substantial trends associated with MOC component are in run 103  Substantial trends associated with non-MOC component in run 104, OC- 12, in subpolar latitudes … Blue = run 202 Red = run 103 Green = run 104 Black = OC-12

23. So what happens to the Subpolar Gyre? Annual mean strength (Sv)  subpolar gyre spins down in all hindcasts, most dramatically in Runs 103, 104 (halving in strength), less so in Run 202, OC-12 (50% weakening)  As observed in 1990s? (Häkkinen & Rhines 2004) Run 202 Run 103 Run 104 OC-12

24. But we must bear in mind OCCAM SST errors: Then got smaller  Got bigger   Errors are Model SST minus NOC SST (ship-based measurements)  Seemingly smallest in the highest resolution experiment  So perhaps we have most faith in this simulation?  But can we do better, e.g., state estimation, new models? …

25. Latest ECCO State Estimation ECCO Near Real-Time Ocean Estimate for April 29, 2008: Temp. anomaly (°C) relative to average seasonal cycle at surface See http://ecco.jpl.nasa.gov/external/index.php

26. And OCCAM being superceded by NEMO… Hindcast now completed for 1958-2001 at 0.25°, e.g., annual-mean SST, 2001: See http://www.noc.soton.ac.uk/nemo/NEMORUNS/nemorun.php

27. Early indications are encouraging - Drifts are small See http://www.noc.soton.ac.uk/nemo/NEMORUNS/nemorun.php Salinity:

28. Early indications are encouraging - Drifts are small See http://www.noc.soton.ac.uk/nemo/NEMORUNS/nemorun.php Temperature:

29. Revealing the longer-term warming annual-mean SST (2001) minus annual-mean (initial state, 1958): See http://www.noc.soton.ac.uk/nemo/NEMORUNS/nemorun.php

30. Along 30°W in the Atlantic: annual-mean SST (2001) minus Levitus climatology: See http://www.noc.soton.ac.uk/nemo/NEMORUNS/nemorun.php

31. Summary  The surface North Atlantic has warmed dramatically since the mid-1980s (+ somewhat more since the 1960s) - seen in observations, well captured in OC-4 hindcast  Surface warming extends to intermediate depths - almost all of the gain in full-depth ocean heat content (OHC) accounted for in upper 1500 m  Pattern of SST & OHC anomalies reveals strongest warming in mid-latitudes (some warming in tropics)  Mid-latitude warming largely controlled by anomalous convergence of ocean heat transport (OHT) in the region: +ve OHT anomaly in northern subtropics, -ve OHT anomaly further north in subpolar gyre  Further developments of OC-4 demonstrate sensitivity of results to “model details” & surface forcing

32. Summary (ctd)  Initial heat budgets with eddy-permitting model hindcast  Subsequent analysis of eddy-resolving model hindcast - preliminary analysis suggests improved realism (SST)  Substantial differences (between eddy-permitting & eddy-resolving) in representation of decadal changes in the “non-MOC” component of OHT  Mechanisms: NAO, AMO, anthropogenic influence ???  Potential for forecasting? (Smith et al., 2007; Keenlyside et al., 2008)  Hindcasting: forcing with “best” surface boundary conditions vs. adjoints & assimilation  Progress continues with NEMO …

33. Papers Marsh, R., de Cuevas, B. A., Coward, A. C., Bryden, H. L., and M. Alvarez (2005). Thermohaline circulation at three key sections in the North Atlantic over 1985-2002. Geophys. Res. Lett., 32, L10604, doi:10.1029/2004GL022281. Marsh. R., de Cuevas, B. A., Coward, A. C., Nurser, A. J. G., and S. A. Josey (2005). Water mass transformation in the North Atlantic over 1985- 2002 simulated in an eddy-permitting model. Ocean Science, 1, 127-144. Marsh, R., Josey, S. A., de Cuevas, B. A., Redbourn, L. J., and G. D. Quartly (2008). Mechanisms for recent warming of the North Atlantic: Insights gained with an eddy-permitting model. J. Geophys. Res., 113, C04031, doi:10.1029/2007JC004096. Marsh, R., de Cuevas, B. A., Coward, A. C., Jacquin, J., Hirschi, J. J.-M., and S. A. Josey. Recent changes in the North Atlantic circulation simulated with eddy-permitting and eddy-resolving ocean models, in prep.